Pradel models and 2 age classes

[shelly]

Hi,
I am investigating the impact of disease on survival rates and population growth rate. I am trying to fit Pradel models to a data set consisting of 2 groups of individuals.
1. Individuals that were first captured as Adults (aged 2+ years)
2. Individuals that were first captured as Subadults (subadults = 1 year olds; so these individuals are Subadults for their first interval and then adults after that. )

My initial CJS models have shown that disease impacts predominately on adult survival.

In order to examine changes in Lambda for the entire adult portion of the population I have combined the adults of both groups into one group (removing the 1st capture for those adults first captured as subadults i.e. 10110100 becomes 00110100). I realise that when fitting the models to this dataset, the estimated recapture rates will be incorrect, so I have modeled recapture rates as either sex-dependent or constant (the best supported p models from my previous CJS analysis) and then fixed these parameters to be the same as those obtained in the CJS analysis. Is this a valid way of overcoming this problem and examining changes in Lambda within the ‘adult’ population?

After reading Chapter 13 and the Nichols et al. Ecology paper I understand that in order to examine changes in Lambda for a population with two age classes as a whole, it is necessary to utilise the reverse time multi-state robust design method they discuss. Is there a guide as to how to implement these models in MARK? Do you reverse the capture histories yourself and then use the closed robust ms option (1000 apologies if this is one of those 'RTFM' answers)? Are there any other options to addressing this problem (with 8 primary samples, at least 28 secondary samples for each, and 3 strata – I just want to explore my options before I start!)?

all advice much appreciated!

[Bill Kendall]

This only addresses part of your question, but in reversing time you only need to reverse the order of your primary periods. Within a primary period you can keep the history in forward time. How important this point is depends on your best fitting models within a primary period. I'm assuming you are using closed models within primary period. If your best fitting model does not involve trap response, I guess it does not matter whether your capture history is in forward or backward time. However, if it does involve trap response, you want to keep those pieces in forward time.

So, if your original history involved sampling times 1-4, where samples 1 and 2 were in primary period 1 and samples 3 and 4 were collected in primary period 2, the original history would follow the following sequence:

1234

For a reverse time analysis, where only the order of primary periods is reversed, the new history would follow the following sequence:

3412


Cheers,
Bill